Adaptive step goals and rewards: a longitudinal growth model of daily steps for a smartphone-based walking intervention

Abstract

Adaptive interventions are an emerging class of behavioral interventions that allow for individualized tailoring of intervention components over time to a person's evolving needs. The purpose of this study was to evaluate an adaptive step goal + reward intervention, grounded in Social Cognitive Theory delivered via a smartphone application (Just Walk), using a mixed modeling approach. Participants (N = 20) were overweight (mean BMI = 33.8 ± 6.82 kg/m), sedentary adults (90% female) interested in participating in a 14-week walking intervention. All participants received a Fitbit Zip that automatically synced with Just Walk to track daily steps. Step goals and expected points were delivered through the app every morning and were designed using a pseudo-random multisine algorithm that was a function of each participant's median baseline steps. Self-report measures were also collected each morning and evening via daily surveys administered through the app. The linear mixed effects model showed that, on average, participants significantly increased their daily steps by 2650 (t = 8.25, p < 0.01) from baseline to intervention completion. A non-linear model with a quadratic time variable indicated an inflection point for increasing steps near the midpoint of the intervention and this effect was significant (t = -247, t = -5.01, p < 0.001). An adaptive step goal + rewards intervention using a smartphone app appears to be a feasible approach for increasing walking behavior in overweight adults. App satisfaction was high and participants enjoyed receiving variable goals each day. Future mHealth studies should consider the use of adaptive step goals + rewards in conjunction with other intervention components for increasing physical activity.

Your Personal Motivator is with You: A Systematic Review of Mobile Phone Applications Aiming at Increasing Physical Activity

“…Four other qualitative studies (of a total of five qualitative studies) [18,56,94,119] provided supportive results, and another qualitative study [98] provided mixed results on PA. For low goals, there were four qualitative and four quantitative studies. All four quantitative studies [115,122,123,137] and four qualitative studies [117,118,126,136] provided supportive effects on PA. Hence, the combined evidence of meta-synthesis and vote counting indicated that goal setting seems to foster PA and that users prefer low over high goals.…”

“…Goals can be (1) app-set, based on users' PA baseline; (2) user-set, with goals set either independently by users or by following researchers' recommendations; or (3) a mix of app-and user-set, where users choose among app-provided predefined goals. Levels of goal challenges are either higher or lower: (1) higher levels mean more challenging goals are formulated, typically in absolute numbers (e.g., taking 10,000 steps/day), or as challenging increases in PA (e.g., 20% weekly increase in steps/day), or as workout sessions planned by a (virtual) trainer (e.g., running activities, 1 km/session) [16,18,56,94,98,119,128]; (2) lower goals are less challenging and are also formulated as an absolute increase in PA (e.g., 10% weekly increase in steps/day) or as a gradual increase [115,117,118,122,123,126,136,137].…”

“…Collectively, users mostly considered high goals as too ambitious, which could be the reason for the non-significant effects in Fanning et al [128]. For users receiving "low" goals, studies mostly reported positive effects on PA [115,122,123,137]. Positive effects were also supported by qualitative studies, where users in interviews or focus groups mentioned that they found these goals convincing and doable [118,126,136].…”

Can the artificial intelligence technique of reinforcement learning use continuously-monitored digital data to optimize treatment for weight loss?

“…These types of changes are an unusual intervention feature within a lifestyle modification program, so participant satisfaction is significant. Previous work has observed that daily novelty in goals is acceptable and efficacious to participants as well (Korinek et al, 2018; Poirier et al, 2016); it may be that novelty in intervention components maintains participant engagement. Participants did express a desire for the automated text messages to be better personalized, so a future iteration of the system might employ more sophisticated tailoring.…”

“…The adaptive intervention was administered on a smartphone application, in which new physical activity goals were continuously assigned and positive reinforcement was provided for meeting goals. While using the application, participants increased their daily steps by over 2600 per day, as compared to the baseline period in which physical activity goals were not specified (Korinek et al, 2018). Additionally, individual differences in predictors of success on a given day support that an individual-level of tailoring in interventions is justified and may further improve outcomes (Phatak et al, 2018).…”